Investigators sought to train and validate a machine learning model to distinguish paper mill publications from genuine articles and assess the prevalence of paper mill publications.